Upgrading Rolls mini mixer MX-28 to get an Hi-Fi sound

[rantony]

Hello everybody,

I have a mini Rolls mixer, model MX-28. It's a very cute and useful product. It's very small, so I can put it on the top of a keyboard to monitor my sound. I like it a lot, but it has a flaw. When I feed it with a bass sound, it won't reproduce every details so some warmness is lost. I would like to mod it to get a better frequency response and thus a more accurate reproduction.



Here is a link to the schematic (it's public):

http://www.rolls.com/pdf/M_MX28.pdf

As you can see, it is using five BA4560 opamps. The 120pF (x8) and the 50nF (x1) capacitors are ceramic. I thought about changing the opamps for better ones, and changing the 120pF ceramic capacitors for mica's. What do you think about this ? Any suggestions for the opamps ?

Another question I have is about the power source: The unit uses 12V. Isn't it a limitation ? Aren't opamps giving better results when they're powered with 18V or 24V ?

Thanks for your help !

[amptramp]

Operating from the 12 VDC supply they recommend, the op amps are biased at 5.67 volts.  This is set by R43 and R44 which gives the base of the power supply transistor 0.55 times the input voltage after the drop at the input protection diode and this is followed by the Vbe drop of the power supply transistor, which is not a regulator - it is a capacitance multiplier that increases C26 by its h_fe.  The 0.55 multiplier is set by R43 and R44.  This should be OK.

The 120 pF caps in conjunction with the 47K feedback resistors give a rolloff at 28.2 KHz, which is OK for value, but ceramics do not belong in audio, especially if they are used in any loud ambient where they may become microphonic.  The ganged level pots at the input have a large effect on frequency response.  When set for maximum gain, they cutoff the low frequencies at 33.86 Hz at the -3db point for the input stage.  The C1 and C2 electrolytic capacitors and the corresponding ones in other channels should be raised in value.

The output stage for J8 and J9 use transistors to boost the op amp output when the op amp output current reaches 23.3 mA.  This is set by the 30 ohm R38 and R41.  At the 5.67 volt bias level, this may not be reached.  Raising this resistor will allow the power output to be handed off to the transistors at a lower level.  If you were to replace the op amps with some of the popular ones like the TL072, which has 192 ohms in series with the output internally, you may never get these transistors to turn on.

The worst problem at this stage is C22 and C24 which set the output bass response.  Going into a zero ohm load, the bass cutoff is 341.4 Hz with this number going down as the load impedance rises.  C19 and C20 are also bad with a cutoff at 159 Hz at low impedances.  Raise the capacitance there for better bass.

The 50 nF ceramic should have a minor effect - it is just there to keep the ground at chassis potential for AC signals, but if it turns microphonic, this will not be good.

This unit is a good starting point for a better mixer (and there are many worse designs out there).  Just do the calculations and set your values where you want them.  I am not familiar with these op amps, but you should determine whether they have the input and output range to handle the signal levels.

[rantony]

Thanks a lot amptramp for your analysis and all the details. It seems that I will have to go through the 'trials and errors' process. I'll try Mica for the 120pF caps. And I'll try a higher value for the 1uF's. But what kind of value do you have in mind for these ? The higher the best ?

The output stage for J8 and J9 use transistors to boost the op amp output when the op amp output current reaches 23.3 mA. This is set by the 30 ohm R38 and R41. At the 5.67 volt bias level, this may not be reached. Raising this resistor will allow the power output to be handed off to the transistors at a lower level. If you were to replace the op amps with some of the popular ones like the TL072, which has 192 ohms in series with the output internally, you may never get these transistors to turn on.

Ok, what you are saying is that it's best to avoid that the transistors turn on ?

The worst problem at this stage is C22 and C24 which set the output bass response. Going into a zero ohm load, the bass cutoff is 341.4 Hz with this number going down as the load impedance rises. C19 and C20 are also bad with a cutoff at 159 Hz at low impedances. Raise the capacitance there for better bass.

Again, which values would you try first ?

Regarding the opamp, I think I'll try to change only the one that handles the signal to the J8 and J9 jacks, as I'm only monitoring my sound with headphones. I might try an OPA2134. Any opinion on this ?

Oh, and by the way, I add the same frequency response issue with a Tapco Mix50 with headphones ...

[amptramp]

If you want hi-fi sound, I would advise going up by a factor of 10 on all electrolytic coupling capacitors.  Remember, the frequencies I have calculated are -3db (half power) points.  This is a substantial drop.  You really want something with no more than 0.5 db drop and the only way to do this is to put the -3db point around a tenth of the lowest frequency you are expecting.  Note that the resistance of the load also affects the frequency response so that higher resistance would reduce the low-frequency turnover (which is what you want).  But if you want a circuit that is compatible with any load, higher capacitance helps.  Note also that most electrolytics have a wide variation in capacitance and going to much larger values assures that there will be less variation from channel to channel both at room temperature and any variations in temperature (and typical electrolytics have a pronounced loss of capacitance both above and below room temperature).

You are not trying to avoid turning transistors Q1 - Q4 on.  These transistors are used as current boosters for the op amps, but you have to be able to drive the base at least one Vbe drop beyond the output in order to get them to supply output current.  What my calculations show is that certain op amps will current limit before they reach an output that would turn the transistors on.  The OPA2134 that you are proposing can source and sink at least 30 mA, so it should work with the existing 30 ohm values for R38 and R41.  BTW, this is a nice amplifier with 8 nV/(Hz)^0.5 noise voltage and wide bandwidth.

The OPA2134 bandwidth may allow you to reduce the value of the 120 pF capacitors on the input stages to permit better high frequency response.  The whole idea of these capacitors is to provide a rolloff before you get to the inherent rolloff of the amplifier and since the OPA2134 is a wideband amp, this could be reduced to a quarter of the 120 pF value (and 30 pF does happen to be a standard value).  Without these capacitors, the input capacitance to ground from the inverting input may be enough to cause a rise in response at high frequncies.  When this response crosses the falling response of the op amp, you get a large phase shift in the feedback which, even if it is not enough to cause oscillation, can cause poor transient response and the possibility of ringing transients that force the amplifier to run out of headroom even at low signal levels.

I would use a socket for the op amps if you intend to experiment with various types as this is a lot better than resoldering a printed circuit board a number of times.

[rantony]

Again, thanks for your nice explanations and advices. I will try to modify only one channel to see if it brings improvements. I will order the required parts and I will do the follow-up once the modifications are done.

Talk to you later !

[quad4]

Would there be any benefits to going to 18, 24 (or even 30 or 36) volts with or without minor mods? Looks like the BA4560 can handle a max VCC of 30v.   The power supply that came with it is 15v.

Is there an easy way to add a clipping indicator for the inputs?  Hoping that the increase in VCC would help the headrooom, but clipping will always happen.

I'm using this at the end of a pedalboard for practice ... mixing in drums, occasional backing tracks ... and going into the effects return of my Mesa Boogie.  Sometimes I use the headphones.  Sometimes I add a Bass as well.  Not  strictly going for hifi and definitely expect compromises for the bass.

Has anyone tried the other mods above?


[Hopefully resurrecting this post from earlier in the decade is a pleasant reminder.  Interesting that the basic design hasn't changed much.  It's a useful mix of of features and form factor and have not seen anything significantly better.  FYI I don't see any transistors in the latest schematic but I do see R38 and R41: https://rolls.com/doc/manuals/manual_MX28.pdf]

[Mark Hammer]

There would likely be a bit of benefit to use of a higher supply voltage.  But the capabilities of the op-amps are not the only concern.  The voltage rating of the electrolytic caps in there should probably be at least 50% higher than the planned supply voltage.  For example, IF one were to use a 30V supply, the next higher standard voltage rating would be 35V, which is rather low from a safety perspective.  You'd want to make sure caps were rated at 50V.  There's a good chance they are already, but do make a point of verifying that.

[Fender3D]

Is there an easy way to add a clipping indicator for the inputs?

+1 on what Mark said about capacitor voltage.

About clipping indicator, no circuit will indicate when output clips, rather it will indicate what LEVEL you set as "clipped".
I mean you build any voltage comparator then feed one input with the rectified signal and the other input with a fixed voltage; if signal is more than the fixed voltage, the LED blinks. You will set the fixed voltage looking at an o-scope when signal clips.
You may build even a simple 1 transistor amplifier going to an LED, but you must check yourself when signal clips, and then make the LED blink...

[quad4]

Thanks for the speedy reply. 

I will upgrade the power supply and then queue up the work on the recto comparator clipping indicator.

[quad4]

Looks like four caps would need to be upgraded.

• 16v: C22, C24, C26
• 25v: C25
• 50v: All the rest

C25 and C26 are in the power stage,  and C22 and C24 are in the headphone out. 

The power supply is 15v so that's not much margin on 16v caps to begin with.   (The OP had a 12v power supply which is a bit more margin, but not 2x.  Suspect the rating on these parts has varied quite a bit over time for this model).  This is in line with a lot of the quality concerns people have had with Rolls in general.

There are a bunch on 120 pF ceramics with no markings beyond 121J. Also C27, a 50nF ceramic cap, from ground to chassis ground. We can ignore the voltage on those, right?

Is it correct that going from 15v to 30v would more/less double the headroom on the input side and give the headphone amp a similar boost?  I think the op amp bias effectively doubles.

[Mark Hammer]

It is always surprising to learn of the voltage-ratings on small ceramic caps.  Equivalent capacitance values in the '60s and '70s would have easily been 20x the size of what one sees now, if not more.  You can expect the brunt of them to have voltage ratings of at least 50V and likely 100V, even if there is no information on them stating so.  I guess when they're that tiny, there's only so much space for printing specs!

And note that not every electrolytic cap needs to have a high voltage rating.  For instance, the inputs on the MX28 have 1uf caps.  What is the likelihood that they will ever see an input signal greater than a couple of volts?  Pretty darn slim.  In contrast, C22 and C24, which you mention, are taking their feed from U5A and U5B, both of which have a reference voltage of around half the supply voltage.  That is, the audio output is "floating" on top of a DC voltage around 15V (assuming a 30V supply), to block that DC reliably from the audio output, you'd want a voltage rating on those caps of 25V (the nearest standard value) or greater.  C27 in the power supply, on the other hand, would need to be in the 50V range if one was going to run a 30V supply; a mere 35V rating would be risky IMHO.

[tubegeek]

For instance, the inputs on the MX28 have 1uf caps.  What is the likelihood that they will ever see an input signal greater than a couple of volts?  Pretty darn slim.

True, but just watch out for anything that might encounter phantom power, that can be an input killer in a worst-case-bonehead scenario. Wrong adapter coming from another mixer input etc. I think that's one rationale for a higher voltage cap on a mixer input. I suppose using that logic I should put "X" line-rated caps on my Speak-On connectors, too, though.

-j, a worst-case bonehead

[Mark Hammer]

Good call.  Not a user of phantom-powered devices, I completely overlooked that.

[amptramp]

If you want to get a good idea of how designers select and derate devices, check out the NASA derating spec:

https://nepp.nasa.gov/docuploads/FFB52B88-36AE-4378-A05B2C084B5EE2CC/EEE-INST-002_add1.pdf

This is a document that covers parts selection, parts qualification (needed for anything going on a spacecraft that does not have an established reliability already) and derating for voltage and temperature.  For example, ceramic and film capacitors have a derating factor of 60% meaning the device can only be exposed to 60% of its rated voltage.  Glass capacitors have a derating of 50%.  Surprisingly, NASA is not the most stringent agency for derating (NAVSEA is a lot more conservative) and its derating factors make sense even for commercial parts.

It is a bit of light reading that shows how a major program sets out standards for it component parts.

[PRR]

> benefits to going to 18, 24 (or even

Rolls is still making this cheap piece of..... uh, I mean low-price bundle of "joy"?

It handles 4V out, more than anything on 1/4" jack should need, and nothing inside should be clipping before the outputs even with crazy high inputs. The op-amps are excellent and 12V isn't hurting them.

Yeah, raise the supply until the caps pop. Fix and raise again.

IMHO, by the time you have "up-rated" this mixer you could have got a pre-owned or new Beringer mixer with more knobs and likely found better "sound".

[tubegeek]

Yeah, raise the supply until the caps pop. Fix and raise again.

It's going to be an Olympic sport one of these years!

IMHO, by the time you have "up-rated" this mixer you could have got a pre-owned or new Beringer mixer with more knobs and likely found better "sound".

USB interface, compression on every channel, all for the price of some kegs of craft beer. The features Behringer is putting into even the cheapest crap most affordable gear they sell kinda boggles my mind. I'd say another good path towards enlightenment in a similar format to the Rolls would be an old Rane FPM-44 half-rack 4-channel mixer - good sound, can be daisy-chained if more channels are wanted, check eBay. Be sure to get the original power supply if you buy a Rane anything - they used an unusual jack (a telco click jack) and 18VAC and you're better off just skipping the "no power supply" deals unless they're dirt cheap - you can't just find the appropriate wall wart in your basement or shed.

https://www.rane.com/pdf/old/fpm44man.pdf

[PRR]

Anything RANE is good. (Except that power supply idea.)

Anything ROLLS is good... for the price, lowest price in the store for the function.

[tubegeek]

(Except that power supply idea.)

I can see how they decided to go with it though. Easy way to do a robust bipolar supply from a wall wart that could work with ANY of their product range, and the telco clicky jack (RJ-?) prevented wrong hook-up disasters. (And since some of their equipment was installed in bars and clubs, wrong hook-up disasters are all but inevitable as it were.)

Inconvenient, sure. But a good technical solution, doncha think?